Rapid Responses to:

LETTERS: Ruud van Winkel and Marc De Hert Reactive hypoglycaemia in severe mental illness BMJ 2009; 338: b2536 [Full text]

Rapid Responses: Submit a response to this article

Rapid Responses published:

A difficult struggle between case study and statistical construction

Zekria Ibrahimi   (30 June 2009)

A difficult struggle between case study and statistical construction 30 June 2009
Zekria Ibrahimi, psychiatric patient Coombs Library UB1 3EU Send response to journal: Re: A difficult struggle between case study and statistical construction	The letter from van Winkel and De Hert provides a statistical fog through which it is not altogether easy to find the medical core. They apply multivariate regression to patients on antipscyhotics in the context of an oral glucose test. They conclude that in a 'control' population of military draftees, hypoglycaemia is found in a minority and would appear asymptomatic(1). They are implying that hypoglycaemia can be a 'normal' state, and that the three cases cited by Suzuki et al are mere irrelevant 'outliers' (2). Under a 'normal' distribution, we would expect an oral glucose test to give figures for a population that would conform to the tediously ubiquitous and dreadfully mathematically complicated 'bell curve'- that is, most at the average, and some at the two extremes. Thus, the 'results' of van Winkel and De Hert are to a degree merely reflecting the normal distribution. Statistics is confined to generalities and abstractions; it is rather similar to a vague haze where we cannot register individuals as individuals. Medicine will not advance without the statistical insight of a van Winkel or De Hert- or without the bright focus on individuals of a Suzuki. The difficult medical problem is that second generation antipsychotics are usually linked to hyperglycaemia. In the three Suzuki cases, some psychiatric drugs were actually causing the reverse, hypoglycaemia. Here was a medical rather than a statistical point. Medicine has traditionally not progressed in the absence of serendipity. To dismiss the Suzuki results because they are unusual would be wrong. Statistical tools are just tools, and cumbersome ones too. So, there is no medical illumination if we do not have both methods- the case study of a Suzuki, or the statistical perspective of a van Winkel and De Hert. REFERENCES: (1) Hypoglycamia in mental illness. Ruud Van Winkel. Marc De Hert. BMJ 2009;338:b2536 (2)Hypoglycaemia induced by second generation antipsychotic agents in schizophrenic non- diabetic patients. Yutaro Suzuki, Junzo Watanabe, Naoki Fukui, Vural Ozdemir, Toshiyuki Someya. BMJ 2009;337:a1792. Competing interests: None declared